A relation between gendered performance differences on standardized tests and in these classes is also suggested by the manner in which ACT scores explain the gendered performance differences we observe. In most classes, adding ACT scores to the model does little to close the gendered performance gap. In these cases whatever caused the gendered performance differences on the ACT does not affect college course performance in the same way. Only when the work done in classes is closely aligned with standardized tests do ACT scores explain the gendered performance differences observed in our classes. This effect is clearest in Math 105 (Precalculus) and Math 115 (Calculus I). According to our LASSO analysis (Supplemental Material), ACT MATH is the best predictor MATH 105 grade, and nearly equal in predictive power to GPAO for MATH 115. This is not true for any of the other STEM courses.